A Fast Image-Spam Filtering System using Support Vector Machine

Abstract

The explosion of Image spam emails has prompted the development of numerous spam filtering techniques. This paper proposes an efficient image spam filtering system using three methods. The first method, File properties, analyses high level features in order to reduce computation cost. The second approach uses Hue, Saturation, Intensity (HSI) color model of histogram and the third method uses Hough line Detection. These three methods filter the image spam by analyzing both images including text and image. The images are collected from three different datasets that are Priceton, Image Spam Hunter and Spam Archieve Datasets. Support Vector Machine (SVM) classifies the input image is spam image or normal image. The experimental result shows the accuracy of different methods on different datasets and evaluates computation time. Among the three methods, Hough line can detect the input image within the minimum processing time required.